| Several related paradigm shifts are taking place in the healthcare field that should enable a profound improvement in the delivery of health care. First is the shift in emphasis away from acute medical treatment, toward comprehensive disease management based on established practice guidelines, treatment algorithms and/or clinical pathways. The goal of disease management is to improve the patient's overall health by preventing the exacerbation of a disease through timely preventive care, thus avoiding the need for expensive and limited acute medical treatment. Second is the movement toward the creation of a next generation computer-based patient record (CPR) that provides access to a patient's lifelong medical data in a cohesive, distributed manner. Currently, a typical patient's medical record is physically scattered among the various sites at which the patient has received treatment. This CPR movement is poised to combine the scattered pieces of a patient's record into a virtual master patient record that will enable the delivery of comprehensive and effective healthcare, prompting the Institute of Medicine to label this next generation CPR as an essential technology for health care.;This emerging environment lays the foundation upon which software systems can be built that could potentially improve a patient's overall health by ensuring that all applicable preventive care activities identified and preformed. Taking advantage of the Internet and the universal access to data that it provides, such systems could search patient records for characteristics and conditions identified by practice guidelines as indicating particular courses of preventive action. Such systems could communicate directly with patients about those actions through the Internet via eMail, telephony, paging, and other means, as they become available. In an attempt to build such a system, the Institute for Artificial Intelligence (IAI) at the George Washington University has developed a prototype Prevention and Wellness WebSite (PWW). The PWW allows consumers to interact with various prototype health care guidelines, identify preventive actions and receive relevant health care support.;The original PWW, though built upon the fundamental concept of intelligent agents, did not meet the requirements of an agent-based system as defined in the literature. In order for a system to be labeled agent-based, the literature requires that its agents exhibit the following characteristics: autonomy, reactivity, pro-activity and social ability. It is the goal of this dissertation to determine the applicability of a proposed hybrid agent architecture to the scenario described above, by extending the functionality of the PWW to satisfy the requirements of agent-based systems. These extensions include inter-agent communication and planning, data awareness, interaction with remote data sources, overall effectiveness, portability, and robustness. A variety of methods were employed in this pursuit, including: proof of concept demonstration implementation, computer-based simulation and scenario demonstration, experimental use of reminder features of the system, and remote porting and installation of the system. As a result of these extensions, the PWW now qualifies as an agent-based system. |
